Add quartiles and IQR to QgsStatisticalSummary calculations

nyalldawson · nyalldawson · commit 31e861123046 · 2015-05-03T20:48:22.000+10:00
diff --git a/python/core/qgsstatisticalsummary.sip b/python/core/qgsstatisticalsummary.sip
@@ -33,6 +33,9 @@ class QgsStatisticalSummary
         Minority, //!< Minority of values
         Majority, //!< Majority of values
         Variety, //!< Variety (count of distinct) values
+        FirstQuartile, //!< First quartile
+        ThirdQuartile, //!< Third quartile
+        InterQuartileRange, //!< Inter quartile range (IQR)
         All
     };
     
@@ -127,6 +130,27 @@ class QgsStatisticalSummary
      * @see minority
      */
     double majority() const;
+    
+        /** Returns the first quartile of the values. The quartile is calculated using the
+     * "Tukey's hinges" method.
+     * @see thirdQuartile
+     * @see interQuartileRange
+     */
+    double firstQuartile() const;
+
+    /** Returns the third quartile of the values. The quartile is calculated using the
+     * "Tukey's hinges" method.
+     * @see firstQuartile
+     * @see interQuartileRange
+     */
+    double thirdQuartile() const;
+
+    /** Returns the inter quartile range of the values. The quartiles are calculated using the
+     * "Tukey's hinges" method.
+     * @see firstQuartile
+     * @see thirdQuartile
+     */
+    double interQuartileRange() const;
 
 };
 
diff --git a/src/core/qgsstatisticalsummary.cpp b/src/core/qgsstatisticalsummary.cpp
@@ -41,6 +41,8 @@ void QgsStatisticalSummary::reset()
   mSampleStdev = 0;
   mMinority = 0;
   mMajority = 0;
+  mFirstQuartile = 0;
+  mThirdQuartile = 0;
   mValueCount.clear();
 }
 
@@ -76,9 +78,13 @@ void QgsStatisticalSummary::calculate( const QList<double> &values )
     mSampleStdev = qPow( sumSquared / ( values.count() - 1 ), 0.5 );
   }
 
-  if ( mStatistics & QgsStatisticalSummary::Median )
+  QList<double> sorted;
+  if ( mStatistics & QgsStatisticalSummary::Median
+       || mStatistics & QgsStatisticalSummary::FirstQuartile
+       || mStatistics & QgsStatisticalSummary::ThirdQuartile
+       || mStatistics & QgsStatisticalSummary::InterQuartileRange )
   {
-    QList<double> sorted = values;
+    sorted = values;
     qSort( sorted.begin(), sorted.end() );
     bool even = ( mCount % 2 ) < 1;
     if ( even )
@@ -91,6 +97,68 @@ void QgsStatisticalSummary::calculate( const QList<double> &values )
     }
   }
 
+  if ( mStatistics & QgsStatisticalSummary::FirstQuartile
+       || mStatistics & QgsStatisticalSummary::InterQuartileRange )
+  {
+    if (( mCount % 2 ) < 1 )
+    {
+      int halfCount = mCount / 2;
+      bool even = ( halfCount % 2 ) < 1;
+      if ( even )
+      {
+        mFirstQuartile = ( sorted[halfCount / 2 - 1] + sorted[halfCount / 2] ) / 2.0;
+      }
+      else //odd
+      {
+        mFirstQuartile = sorted[( halfCount  + 1 ) / 2 - 1];
+      }
+    }
+    else
+    {
+      int halfCount = mCount / 2 + 1;
+      bool even = ( halfCount % 2 ) < 1;
+      if ( even )
+      {
+        mFirstQuartile = ( sorted[halfCount / 2 - 1] + sorted[halfCount / 2] ) / 2.0;
+      }
+      else //odd
+      {
+        mFirstQuartile = sorted[( halfCount  + 1 ) / 2 - 1];
+      }
+    }
+  }
+
+  if ( mStatistics & QgsStatisticalSummary::ThirdQuartile
+       || mStatistics & QgsStatisticalSummary::InterQuartileRange )
+  {
+    if (( mCount % 2 ) < 1 )
+    {
+      int halfCount = mCount / 2;
+      bool even = ( halfCount % 2 ) < 1;
+      if ( even )
+      {
+        mThirdQuartile = ( sorted[ halfCount + halfCount / 2 - 1] + sorted[ halfCount + halfCount / 2] ) / 2.0;
+      }
+      else //odd
+      {
+        mThirdQuartile = sorted[( halfCount + 1 ) / 2 - 1 + halfCount ];
+      }
+    }
+    else
+    {
+      int halfCount = mCount / 2 + 1;
+      bool even = ( halfCount % 2 ) < 1;
+      if ( even )
+      {
+        mThirdQuartile = ( sorted[ halfCount + halfCount / 2 - 2 ] + sorted[ halfCount + halfCount / 2 - 1 ] ) / 2.0;
+      }
+      else //odd
+      {
+        mThirdQuartile = sorted[( halfCount + 1 ) / 2 - 2 + halfCount ];
+      }
+    }
+  }
+
   if ( mStatistics & QgsStatisticalSummary::Minority || mStatistics & QgsStatisticalSummary::Majority )
   {
     QList<int> valueCounts = mValueCount.values();
diff --git a/src/core/qgsstatisticalsummary.h b/src/core/qgsstatisticalsummary.h
@@ -49,7 +49,10 @@ class CORE_EXPORT QgsStatisticalSummary
       Minority = 512, //!< Minority of values
       Majority = 1024, //!< Majority of values
       Variety = 2048, //!< Variety (count of distinct) values
-      All = Count | Sum | Mean | Median | StDev | Max | Min | Range | Minority | Majority | Variety
+      FirstQuartile = 4096, //!< First quartile
+      ThirdQuartile = 8192, //!< Third quartile
+      InterQuartileRange = 16384, //!< Inter quartile range (IQR)
+      All = Count | Sum | Mean | Median | StDev | Max | Min | Range | Minority | Majority | Variety | FirstQuartile | ThirdQuartile | InterQuartileRange
     };
     Q_DECLARE_FLAGS( Statistics, Statistic )
 
@@ -143,6 +146,27 @@ class CORE_EXPORT QgsStatisticalSummary
      */
     double majority() const { return mMajority; }
 
+    /** Returns the first quartile of the values. The quartile is calculated using the
+     * "Tukey's hinges" method.
+     * @see thirdQuartile
+     * @see interQuartileRange
+     */
+    double firstQuartile() const { return mFirstQuartile; }
+
+    /** Returns the third quartile of the values. The quartile is calculated using the
+     * "Tukey's hinges" method.
+     * @see firstQuartile
+     * @see interQuartileRange
+     */
+    double thirdQuartile() const { return mThirdQuartile; }
+
+    /** Returns the inter quartile range of the values. The quartiles are calculated using the
+     * "Tukey's hinges" method.
+     * @see firstQuartile
+     * @see thirdQuartile
+     */
+    double interQuartileRange() const { return mThirdQuartile - mFirstQuartile; }
+
   private:
 
     Statistics mStatistics;
@@ -157,6 +181,8 @@ class CORE_EXPORT QgsStatisticalSummary
     double mSampleStdev;
     double mMinority;
     double mMajority;
+    double mFirstQuartile;
+    double mThirdQuartile;
     QMap< double, int > mValueCount;
 };
 
diff --git a/tests/src/core/testqgsstatisticalsummary.cpp b/tests/src/core/testqgsstatisticalsummary.cpp
@@ -84,6 +84,39 @@ void TestQgsStatisticSummary::stats()
   QCOMPARE( s.variety(), 7 );
   QCOMPARE( s.minority(), 3.0 );
   QCOMPARE( s.majority(), 12.0 );
+
+  //test quartiles. lots of possibilities here, involving odd/even/divisible by 4 counts
+  values.clear();
+  values << 7 << 15 << 36 << 39 << 40 << 41;
+  s.calculate( values );
+  QCOMPARE( s.median(), 37.5 );
+  QCOMPARE( s.firstQuartile(), 15.0 );
+  QCOMPARE( s.thirdQuartile(), 40.0 );
+  QCOMPARE( s.interQuartileRange(), 25.0 );
+
+  values.clear();
+  values << 7 << 15 << 36 << 39 << 40 << 41 << 43 << 49;
+  s.calculate( values );
+  QCOMPARE( s.median(), 39.5 );
+  QCOMPARE( s.firstQuartile(), 25.5 );
+  QCOMPARE( s.thirdQuartile(), 42.0 );
+  QCOMPARE( s.interQuartileRange(), 16.5 );
+
+  values.clear();
+  values << 6 << 7 << 15 << 36 << 39 << 40 << 41 << 42 << 43 << 47 << 49;
+  s.calculate( values );
+  QCOMPARE( s.median(), 40.0 );
+  QCOMPARE( s.firstQuartile(), 25.5 );
+  QCOMPARE( s.thirdQuartile(), 42.5 );
+  QCOMPARE( s.interQuartileRange(), 17.0 );
+
+  values.clear();
+  values << 6 << 7 << 15 << 36 << 39 << 40 << 41 << 42 << 43 << 47 << 49 << 50 << 58;
+  s.calculate( values );
+  QCOMPARE( s.median(), 41.0 );
+  QCOMPARE( s.firstQuartile(), 36.0 );
+  QCOMPARE( s.thirdQuartile(), 47.0 );
+  QCOMPARE( s.interQuartileRange(), 11.0 );
 }
 
 QTEST_MAIN( TestQgsStatisticSummary )
